I am trying to add vector KNN support for my data to feed into an AI system.
I am currently using only plain tables with sources and indexer defined in the .conf file (no data_dir obviously).
I am trying to create an rt-table in rt-mode (via sql) but I get the “CREATE TABLE requires data_dir to be set in the config file“ so no luck there, but is there a way to make plain and rt tables coexist ?
Do I need to define the rt table in plain mode (that is using the conf file) ?
Can anyone share an example of how to achieve this if possible?
Or do I need to convert all my tables to RT in order to use the vector attribute ?
Between rt- or plain-mode and rt- or plain tables it gets quite confusing at times ..
Thanks for confirming, where do I find the datafiles for the rt table defined in plain mode, I checked /var/lib/manticore/ (where all the plain indices are) but I can only find the .meta file there. Do I need to define a data_dir somewhere ?
Update
The whole issue that originated the question (I am new to rt tables …) is that when you create an rt-table in plain mode (with the conf file) deleting the table is not that straight forward, you need to:
stop the manticore server
delete all rt-table files (.meta .ram and any .sp if there are any)
restart the server
If you forget to delete any of these you may find some strange behaviours (like the table gets created , maybe,but it is not listed in the db ..)